Objective Quality and Intelligibility Prediction for Users of Assistive Listening Devices: Advantages and limitations of existing tools

Abstract

This article presents an overview of twelve existing objective speech quality and intelligibility prediction tools. Two classes of algorithms are presented, namely intrusive and non-intrusive, with the former requiring the use of a reference signal, while the latter does not. Investigated metrics include both those developed for normal hearing listeners, as well as those tailored particularly for hearing impaired (HI) listeners who are users of assistive listening devices (i.e., hearing aids, HAs, and cochlear implants, CIs). Representative examples of those optimized for HI listeners include the speech-to-reverberation modulation energy ratio, tailored to hearing aids (SRMR-HA) and to cochlear implants (SRMR-CI); the modulation spectrum area (ModA); the hearing aid speech quality (HASQI) and perception indices (HASPI); and the PErception MOdel - hearing impairment quality (PEMO-Q-HI). The objective metrics are tested on three subjectively-rated speech datasets covering reverberation-alone, noise-alone, and reverberation-plus-noise degradation conditions, as well as degradations resultant from nonlinear frequency compression and different speech enhancement strategies. The advantages and limitations of each measure are highlighted and recommendations are given for suggested uses of the different tools under specific environmental and processing conditions.

Non-intrusive intelligibility prediction using a codebook-based approach

“…Understanding speech becomes a challenging task in adverse listening conditions such as "the cocktail party scenario" especially for hearing impaired individuals [1], [2]. Speech enhancement algorithms aim to improve speech intelligibility for hearing aid users [3], [4], [5]. However, speech enhancement algorithms may be beneficial in some acoustic scenarios whereas the same algorithms can have a negative impact on quality and intelligibility in other conditions [5], [6].…”

“…However, speech enhancement algorithms may be beneficial in some acoustic scenarios whereas the same algorithms can have a negative impact on quality and intelligibility in other conditions [5], [6]. Thus, it would be beneficial for HA users if speech enhancement algorithms are automatically limited to scenarios in which they provide an improvement in speech intelligibility [3], [4]. This could be facilitated by an objective speech intelligibility metric processed online in the HA.…”

Subjective and Objective Sound-Quality Evaluation of Adaptive Feedback Cancellation Algorithms

“…Five commonly-used performance criteria, chosen based on the analysis in [61], were used to evaluate the predictive capabilities of the objective measures.…”

“…The chosen statistical test for the correlation-based criteria was the Fisher transformation z-test [61]; for the RMSE-based criteria, the test suggested in [61], based on the test proposed in [63], was chosen, i. e.:…”

“…In order to quantify these aspects, the results in terms of the different performance criteria, defined in Section V, are shown in Table III. The best performances, separately calculated for speech and music signals, are reported in bold (up to statistical indifference) for two performance criteria, as done in [61], i. e. r sig and ε-RMSE. The cases that yielded a value of ε-RMSE = 0 are indicated by '-'.…”